Managing Assembly Production Processes with KANBAN and CONWIP

To control the production of parts in a production process, managers can choose aproper production control policy. This paper provides a performance analysis ofproduction control systems in a business process, which is modeled as adiscrete-event system and depicted by AID (Activity Interaction Diagram).KANBAN and CONWIP controls are focused on and analyzed. The periodicbehavior of a token transaction system and the concepts of critical circuit andtokens are used. When a business process behaves periodically, Little’s law can beused to calculate the cycle time, inventory, and throughput of the system. By usingthe theory of token transaction systems, and employing the law, we show howminimum WIP (Work-In-Process) of a system can be calculated that allows thesystem to have maximum possible throughput. As an application of the theory, weprovide a performance comparison between KANBAN and CONWIP. The resultsshow that there is no general superiority between KANBAN and CONWIP.Appropriate design of the whole system decides which one is superior in certainsituations